CaMKII Control of Actin Cytoskeletal Dynamics

Abstract

The calcium calmodulin dependent kinase (CaMKII) binds and organizes actin Binding is abolished by calcium calmodulin, triggering cytoskeletal remodeling with important implications for synaptic transmission in dendritic spines; CaMKII and actin-rich post-synaptic micro-compartments that expand within one minute upon synaptic stimulation to initiate long-term changes associated with learning. CaMKII organizes static F-actin bundles in-vitro, but fluorescence microscopy single particle tracking (SPT) of GFP-tagged CaMKII holoenzymes (GFP-CaMKII) in spines has not detected stimulus-dependent jumps in mobility consistent with un-bundling. However, changes in binding affinities, in response to osmotic and mechanical forces, might provoke diverse cytoskeletal CaMKII-actin architectures in vivo. To look for these, we initially used SPT in live cells and found both rat neuronal CaMKII isoforms bound RFP-actin labeled stress fibers weakly, comparable to G-actin, with strong binding to the fibers obtained only when multiple holoenzyme subunits are engaged (Khan et al. 2016. Biophys. J.111,395-408). We have now measured the binding of GFP-CaMKII molecules to rhodamine-phalloidin labeled, single F-actin filaments and characterized the architectural transitions of the formed filament networks to mechano-osmotic stress. CaMKII induced formation of mechanically-resilient networks with aster-like nodes. Cryo-electron tomography showed CaMKII localized in small clusters (n < 5) at the nodes. Although nodes disappeared rapidly upon addition of calcium calmodulin, the network structure took several minutes to dissipate. Both C. elegans and human CaMKIIs formed networks at stoichiometric concentrations, revealing an evolutionary-conserved interaction. At the micromolar concentrations investigated, CaMKII-actin bundles were sparsely distributed in the networks. Macromolecular crowding by addition of polyethylene glycol or force generated by addition of myosin motors plus ATP bundled the filaments; but CaMKII retarded, rather than assisted, this transition. CaMKII also protected against filament fragmentation by motor forces. We propose that stimulation-triggered disruption of spine CaMKII-actin interactions has two distinct effects, formation of a compliant F-actin network primarily due to loss of nodal contacts and polymerization of the released G-actin. We are simulating these processes in a 3-dimensional “virtual” spine to determine CaMKII, G and F-actin concentrations and affinities that best mimic the observed spine dynamics upon stimulation.